Can someone break down the input lag of an LCD vs CRT?

[Vantavia]

The refresh cycle of a CRT seems to be pretty much instant (as far as I know).

But LCD has -
an intial delay from processing
a delay from pixel transition based on colour different 1-10ms
and a delay for the refresh cycle which makes the image at the bottom of the screen like 8ms behind
with some overlap or something which brings the total a little bit lower *shrugs*

[spacediver]

a CRT running at 100 hz will render the first pixel in the top row about 10 ms before it renders the last pixel in the bottom row.

[Glide]

and a delay for the refresh cycle which makes the image at the bottom of the screen like 8ms behind

Most displays have this. And it is not "like 8ms" it is the duration of the scanout.
At 60Hz that is 16.67ms (1000/60)
At 120Hz that would be 8.33ms (1000/120)

LCD, CRT, OLED all do this.
Plasma TVs do not.
The OLED displays used in VR headsets are supposed to be "global displays" which do not do this either.

I suspect that they must buffer the entire frame and then update the display in <1ms
This would actually be higher latency, but avoids the skew that "rolling displays" have with horizontal motion.

[spacediver]

LCD, CRT, OLED all do this.
Plasma TVs do not.

I'm pretty sure Plasma TV's use progressive scanning. Where did you hear otherwise? I've heard of some global display LCD prototypes that update all pixels at once, but not plasma.

[Glide]

I'm pretty sure Plasma TV's use progressive scanning. Where did you hear otherwise? I've heard of some global display LCD prototypes that update all pixels at once, but not plasma.

If you measure a plasma TV with the Leo Bodnar lag tester, you get the same latency for all positions.
I assume it is required for the way they drive the display using multiple subfields.
I would not be surprised if DLP behaved the same way.

[spacediver]

Yes, according to Michael Abrash, some DLPs are global displays.

http://blogs.valvesoftware.com/abrash/r ... s-the-eye/

About the leo bodnar tests, is that the case for all plasmas, or just some?

[Glide]

About the leo bodnar tests, is that the case for all plasmas, or just some?

All that I have seen. Though I would assume all, due to the fact that they draw everything using sub-fields - which is the reason that I expected DLP to update globally as well.

[Vantavia]

So what IS the full process/chain for LCD and CRT?
e.g.

a. blah 2 ms
b. blah 8.33 ms
c. blah 2-8.33 ms
d. blah 2 ms
>repeat

total input lag: blah-blah

There are so many terms for what appear to be the same thing it's hard not to add more things in the chain by accident.

[spacediver]

well flood's input lag measurements on his CRT show that from the moment the mouse USB poll is registered to the moment the phosphors light up is about 0.45 ms. I imagine most of that happens before the tube receives the VGA signal.

That's about as much as I know.

(here is the post about his measurements)


http://www.esreality.com/post/2691945/m ... surements/

also see

http://forums.blurbusters.com/viewtopic.php?f=10&t=1381

[Vantavia]

So now we just need the equivilent measurement for a BenQ monitor @ 120hz and Nvidia vs AMD since Nvidia's GPU scaling seems broken AF.